Electrical current uniformity is critical to the performance and reliability of light emitting diodes (LEDs). By optimizing current spreading uniformly across an LED device, losses due to current crowding in parts of the device are eliminated, efficiency is enhanced, and light extraction is maximized. Thermal and optical performance characteristics may also be optimized to further improve the operation and efficiency of the LED device.
Without the benefit of robust modeling technology, LED chip designers often design, manufacture, and characterize actual LED devices before optimizing the devices. After building, testing and characterizing the actual devices, the designers attempt to determine what changes to implement to best improve device performance. To test the improvement theories of the designers, the entire process is repeated, including build, manufacture and test. This approach may be a very costly and slow process. Typically, a manufacturing run may take two months and significant man-hours to complete.
To improve design characteristics and overall performance of the devices, software simulations may be implemented. Circuit simulation packages, such as SPICE, have been predominant in the industry for nearly two decades. Designers implement simulation packages, such as SPICE, by writing a schematic description or “netlist” (e.g., “SPICE deck”) of the circuit and feeding the schematic description into SPICE for batch mode simulation. Using physics as the basis for the model, SPICE decks are typically written manually by the designer. Because of the difficulties associated with the manual process, SPICE decks are often minimized and simplified. As can be appreciated by those having ordinary skill in the art, a SPICE deck containing 1000 lines of code is generally considered large. A manually written SPICE netlist having less than 1000 lines of code, generally corresponds to less robust modeling. As can be appreciated, the manual process of writing netlists for circuit simulation may be slow and tedious. Further, physics modeling paradigms, such as SPICE, may not provide the capabilities for modeling all of the variables desirable for fully optimizing the device.